Semiconductor packaging and its data transmission methods

By using hybrid bonding pads to directly connect the die and the silicon interposer, high-speed data signals can be transmitted in parallel, solving the problems of limited microbump interconnect density and circuit complexity, and realizing high-bandwidth parallel interconnect and simplified circuit structure.

CN123124834APending Publication Date: 2026-11-17ZHIHAOTONG (TIANJIN) INFORMATION TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611619330.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-20
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In existing 2.5D/3D packaging solutions, the microbump interconnect density between the die and the silicon interposer is limited, and the reliance on serial links for high-speed data transmission leads to complex physical layer circuitry, making it difficult to meet the requirements for high-bandwidth parallel interconnection.

Method used

Hybrid bonding pads are used to directly connect the die to the silicon interposer, achieving high-density interconnects below 10μm. Data transmission links are set up within each die to transmit high-speed data signals in parallel, simplifying the physical layer circuit structure.

Benefits of technology

It increases interconnect density, simplifies the physical layer circuit structure, reduces circuit complexity, reduces the cost of serialization, and improves data transmission efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123124834A_ABST
    Figure CN123124834A_ABST
Patent Text Reader

Abstract

This application provides a semiconductor package and its data transmission method, relating to the field of semiconductor packaging technology. It facilitates increasing the die interconnect density within the semiconductor package and reducing the complexity of the physical layer circuit structure. The package includes: a packaging substrate, a silicon interposer layer located above the packaging substrate, and at least two dies. The silicon interposer layer is connected to the packaging substrate via a bump array, and the at least two dies are connected to the upper surface of the silicon interposer layer via hybrid bonding pads. High-speed data signals, power signals, ground signals, and control signals between the dies and the silicon interposer layer are directly bonded and interconnected via the hybrid bonding pads. Each die contains a data transmission link, which includes: a transmit FIFO queue, a transmit circuit, a receive circuit, and a receive FIFO queue. The high-speed data signals are transmitted in parallel between the transmit and receive circuits of the two dies via the data transmission link.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of semiconductor device packaging technology, and in particular to a semiconductor packaging and data transmission method thereof. Background Technology

[0002] In the field of advanced integrated circuit packaging, high-speed interconnection between multiple dies is crucial for improving system bandwidth and reducing power consumption. In existing 2.5D / 3D packaging solutions, electrical connections between the die and the silicon interposer are typically achieved using micro-bumps. However, due to limitations in bump manufacturing processes and the physical limits of underfill, the micro-bump pitch is usually not below 10μm, with actual mass production typically using 40–60μm. This creates a hard upper limit on interconnect density, making it difficult to meet the demands of high-bandwidth parallel interconnection.

[0003] In addition, the lateral transmission distance between the chips is as long as several millimeters to tens of millimeters, resulting in severe signal attenuation. Existing solutions generally rely on serializers / deserializers (SerDes), clock data recovery (CDR), and equalization circuits to transmit data serially, which leads to a complex physical layer circuit structure and a long end-to-end transmission delay. Summary of the Invention

[0004] In view of this, embodiments of this application provide a semiconductor package and its data transmission method to solve the technical problems in existing multi-die interconnect packaging structures, where the interconnect density is limited due to the use of micro-bump interconnects between the die and the silicon interposer, and the high-speed data transmission between dies relies on serial links, resulting in complex physical layer circuits. This facilitates the improvement of die interconnect density inside the semiconductor package and reduces the complexity of the physical layer circuit structure.

[0005] In a first aspect, embodiments of this application provide a semiconductor package, comprising: a packaging substrate, a silicon interposer located above the packaging substrate, and at least two dies; the silicon interposer is connected to the packaging substrate via a bump array, and the at least two dies are connected to the upper surface of the silicon interposer via hybrid bonding pads; high-speed data signals, power signals, ground signals, and control signals between the dies and the silicon interposer are directly bonded and interconnected via hybrid bonding pads; each die is provided with a data transmission link, the data transmission link comprising: a transmit first-in-first-out queue, a transmit circuit, a receive circuit, and a receive first-in-first-out queue, the high-speed data signals being transmitted in parallel between the transmit circuit and the receive circuit of the two dies via the data transmission link.

[0006] According to one embodiment of this application, the transmitting circuit includes a main drive branch, a first pre-emphasis branch, a first programmable resistor, a second programmable resistor, a first operational amplifier, a second operational amplifier, a first load transistor, a second load transistor, and a feedback resistor; the output terminal of the main drive branch is connected to the output terminal of the first pre-emphasis branch via the first programmable resistor, and the output terminal of the first pre-emphasis branch is connected to the output terminal of the transmitting circuit via the second programmable resistor; the non-inverting input terminal of the first operational amplifier is used to connect to a first reference voltage, the inverting input terminal of the first operational amplifier is connected to the output terminal of the transmitting circuit via the feedback resistor, and the output terminal of the first operational amplifier is connected to the gate of the first load transistor; the non-inverting input terminal of the second operational amplifier is used to connect to a second reference voltage, the inverting input terminal of the second operational amplifier is connected to the output terminal of the transmitting circuit, and the output terminal of the second operational amplifier is connected to the gate of the second load transistor; the first load transistor is connected between the power supply terminal and the first common source node of the main drive branch and the first pre-emphasis branch, and the second load transistor is connected between the second common source node of the main drive branch and the first pre-emphasis branch and the ground terminal.

[0007] According to one embodiment of this application, the main driving branch includes: a first PMOS transistor and a first NMOS transistor, wherein the drains of the first PMOS transistor and the first NMOS transistor are connected together, and the gates of the first PMOS transistor and the first NMOS transistor are connected together to the data input terminal; the first pre-emphasis branch is composed of a second PMOS transistor and a second NMOS transistor; the drains of the second PMOS transistor and the second NMOS transistor are connected together; and the gates of the second PMOS transistor and the second NMOS transistor are connected together to the control terminal of the first pre-emphasis branch.

[0008] According to one embodiment of this application, the first programmable resistor and the second programmable resistor are digitally programmable adjustable resistors. The first programmable resistor and the second programmable resistor each include a plurality of resistor units connected in parallel. Each resistor unit is connected in series with a switching transistor. The switching transistor is controlled to turn on or off by a register configuration signal, and is used to switch the corresponding resistor unit on or off.

[0009] According to one embodiment of this application, the control terminal of the first pre-emphasis branch is configured to control the second PMOS transistor and the second NMOS transistor to turn on when a data transition edge arrives, and to turn off when the data is in a steady state; the signal output by the main drive branch is transmitted to the output terminal of the transmitting circuit via the first programmable resistor and the second programmable resistor, and the signal output by the first pre-emphasis branch is transmitted to the output terminal of the transmitting circuit via the second programmable resistor, for adjusting the pre-emphasis signal enhancement amplitude at the output terminal of the transmitting circuit at the data transition edge; the pre-emphasis signal enhancement amplitude is determined by the resistance ratio of the first programmable resistor to the second programmable resistor.

[0010] According to one embodiment of this application, the first operational amplifier is further configured to adjust the on-resistance of the first load transistor via negative feedback, so that the high level of the output terminal of the transmitting circuit is equal to the first reference voltage; the second operational amplifier is further configured to adjust the on-resistance of the second load transistor via negative feedback, so that the low level of the output terminal of the transmitting circuit is equal to the second reference voltage; wherein, the output swing of the transmitting circuit is the difference between the high level and the low level, and the output swing is adjusted by setting the first reference voltage and the second reference voltage.

[0011] According to one embodiment of this application, the high-speed data signal enters the silicon interposer vertically through the hybrid bonding pads, is transmitted through the lateral interconnect pathways inside the silicon interposer to the hybrid bonding pads below the adjacent die, and is then transmitted vertically to the adjacent die; the power signal and the ground signal are led out to the packaging substrate through the through-silicon vias and the bump array in the silicon interposer.

[0012] According to one embodiment of this application, the transmitting circuit further includes at least one second pre-emphasis branch, and the at least one second pre-emphasis branch is connected to the output terminal of the transmitting circuit via a corresponding third programmable resistor.

[0013] According to one embodiment of this application, the transmitting circuit further includes an eye diagram monitoring circuit or a bit error detection circuit, used to adjust the pre-emphasis signal enhancement amplitude and output swing based on the signal quality feedback from the receiving end.

[0014] Secondly, embodiments of this application also provide a data transmission method for a semiconductor package, comprising: interconnecting high-speed data signals, power signals, ground signals, and control signals between the upper surface of a silicon interposer and at least two dies via hybrid bonding pads; transmitting the high-speed data signals in parallel between the two dies via a data transmission link within each die, and transmitting the control signals via the hybrid bonding pads; and leading the power signals and the ground signals to a package substrate via through-silicon vias and a bump array within the silicon interposer.

[0015] According to one embodiment of this application, the high-speed data signal is transmitted in parallel between two dies via a data transmission link within each die, comprising: outputting a signal from a main drive branch, transmitting the output signal of the main drive branch to the output terminal of a first pre-emphasis branch via a first programmable resistor, transmitting the output signal of the first pre-emphasis branch to the output terminal of a transmitting circuit via a second programmable resistor; adjusting the on-resistance of a first load transistor via a first operational amplifier, and adjusting the on-resistance of a second load transistor via a second operational amplifier.

[0016] According to one embodiment of this application, the step of outputting a signal through the main drive branch includes: when a data transition edge arrives, controlling the first pre-emphasis branch to be turned on, so that the output signal of the first pre-emphasis branch is transmitted to the output terminal of the transmitting circuit via the second programmable resistor; when the data is in a steady state, controlling the first pre-emphasis branch to be turned off, so that the output signal of the main drive branch is transmitted to the output terminal of the transmitting circuit via the first programmable resistor and the second programmable resistor; and adjusting the pre-emphasis signal enhancement amplitude at the output terminal of the transmitting circuit at the data transition edge by adjusting the resistance ratio of the first programmable resistor and the second programmable resistor.

[0017] According to one embodiment of this application, adjusting the on-resistance of the first load transistor via the first operational amplifier includes: adjusting the on-resistance of the first load transistor via negative feedback through the first operational amplifier, so that the high level at the output of the transmitting circuit is equal to the first reference voltage; adjusting the on-resistance of the second load transistor via the second operational amplifier includes: adjusting the on-resistance of the second load transistor via negative feedback through the second operational amplifier, so that the low level at the output of the transmitting circuit is equal to the second reference voltage; and adjusting the output swing of the transmitting circuit by configuring the first reference voltage and the second reference voltage.

[0018] According to one embodiment of this application, the high-speed data signal interconnected between the upper surface of the silicon interposer and at least two dies via the hybrid bonding pad includes: vertically transmitting the high-speed data signal into the silicon interposer via the hybrid bonding pad; transmitting the high-speed data signal to the hybrid bonding pad below the adjacent die via the lateral interconnect path inside the silicon interposer; and then vertically transmitting the high-speed data signal to the adjacent die via the hybrid bonding pad.

[0019] The semiconductor packaging and data transmission method provided in this application improve the multi-die interconnect packaging structure by directly bonding and interconnecting high-speed data signals, power signals, ground signals, and control signals between the die and the silicon interposer via hybrid bonding pads. This improves interconnection density and unifies the bonding interface. Each die includes a transmit FIFO queue, a transmit circuit, a receive circuit, and a receive FIFO queue. High-speed data signals are transmitted between dies in parallel via the aforementioned data transmission links, which simplifies the physical layer circuit structure and reduces the circuit complexity caused by serialization.

[0020] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this specification. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the structure of a semiconductor package provided in an embodiment of this application; Figure 2 This is an overall schematic diagram of the semiconductor packaged transmitting circuit provided in an embodiment of this application; Figure 3 This is a schematic diagram of the current flow in the transmitting circuit during the first operating stage in an embodiment of this application. Figure 4 This is a schematic diagram of the current flow of the transmitting circuit in the second operating stage in an embodiment of this application; Figure 5 This is a schematic diagram of the current flow of the transmitting circuit in the third operating stage in an embodiment of this application; Figure 6 This is a schematic diagram of the current flow of the transmitting circuit in the fourth operating stage in an embodiment of this application; Figure 7 The transient simulation waveforms of the data input terminal, the pre-emphasis control terminal input, and the transmission circuit output corresponding to each working stage in the embodiments of this application are shown. Figure 8 This is a schematic flowchart of a data transmission method for a semiconductor package provided in an embodiment of this application. Detailed Implementation

[0022] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to limit the application. The singular forms "a," "the," and "the" used in this application are also intended to include the plural forms unless the context clearly indicates otherwise. The terms "first," "second," "third," etc., may be used in this application to describe various information or structural modules for the purpose of more clearly describing the scheme, and should not be construed as indicating or implying relative importance or implicitly specifying the number, order, or position of the indicated technical features. Thus, a feature defined with "first," "second," "third," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more; "if" can be interpreted as "when," "when," or "in response to a determination." In this application, "and / or" is used to describe the relationship between related objects, indicating that three relationships may exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural.

[0023] In this specification, unless otherwise expressly specified and limited, the term "connection" should be interpreted broadly. For example, "connection" can mean a direct connection or an indirect connection through an intermediate medium. Furthermore, the term "coupled connection" can mean a direct electrical connection or an indirect electrical connection through an intermediate medium.

[0024] The embodiments described in this specification will now be described in detail. It should be understood that the specific embodiments described herein are only for the purpose of aiding in understanding this application.

[0025] Example 1 In the field of advanced integrated circuit packaging, multiple dies are interconnected at high density via silicon interposers. Electrical connections between the dies and the silicon interposers are typically achieved using micro-bumps. However, due to limitations in bump manufacturing processes and the physical constraints of underfill, the interconnect density of micro-bumps has an upper limit, making it difficult to meet the demands of high-bandwidth parallel interconnects.

[0026] Furthermore, the long lateral transmission channels between chips result in significant signal attenuation. Existing solutions typically integrate serializer / deserializer (SerDes), clock data recovery (CDR), and equalization circuits at both the transmitting and receiving ends to transmit data serially. This leads to a complex physical layer circuit structure, requiring the integration of multiple stages of processing, including serialization, encoding, equalization, CDR recovery, decoding, and deserialization. This not only occupies a large chip area but also results in high end-to-end transmission delays.

[0027] See Figure 1This application provides a semiconductor package including a package substrate, a silicon interposer located above the package substrate, and at least two dies. The silicon interposer is connected to the package substrate via a bump array, and the at least two dies are directly bonded to the upper surface of the silicon interposer via hybrid bonding (HB) pads.

[0028] In this process, high-speed data signals, power signals, ground signals, and control signals between the die and the silicon interposer are directly bonded and interconnected via hybrid bonding pads. By replacing the traditional microbump (μBump) connection with a hybrid bonding connection, high-density interconnects below 10μm can be achieved, and data signals and high-current power / ground signals can be carried. Optionally, the hybrid bonding pads are metal pads, preferably copper pads.

[0029] Understandably, in existing solutions, high-speed data signals are typically interconnected between the die and the silicon interposer via microbumps, while power and ground signals require additional bumps or wiring. In this embodiment, however, high-speed data signals, power signals, ground signals, and control signals are all interconnected through the same hybrid bonding interface, achieving high-density interconnects below 10μm. Furthermore, it eliminates the need for separate interconnect structures for different signal types, improving the utilization of the bonding interface and simplifying the packaging process.

[0030] Each die contains a data transmission link, which includes a first-in-first-out (FIFO) transmission queue, a transmission circuit, a receiving circuit, and a first-in-first-out (FIFO) receiving queue. The high-speed data signal is transmitted in parallel between the transmission circuits and receiving circuits of the two dies via the data transmission link.

[0031] Specifically, the data generated inside die A is written into the transmitting FIFO. The transmitting FIFO achieves cross-clock domain isolation, decoupling the internal logic clock of the die from the interconnect transmission clock, while providing data buffering to smooth data rate fluctuations. The data output from the transmitting FIFO is a voltage signal that descends vertically through the hybrid bonding pads into the silicon interposer, is transmitted through the lateral interconnect path inside the silicon interposer to the bottom of the opposite die B, and then ascends vertically through the hybrid bonding pads to reach the opposite die B.

[0032] The receiving circuit of the peer chip B can be an adjustable threshold CMOS amplifier or Schmitt trigger. The corresponding judgment threshold is configured according to the output swing of the transmitting circuit. It directly receives the voltage signal without clock data recovery (CDR). The received data is written into the receiving end FIFO and then read by the internal logic circuit of chip B.

[0033] It should be noted that the transmitting FIFO writes data using the internal logic clock of the transmitting chip and reads data using the transmission clock; the receiving FIFO writes data using the receiving clock and reads data using the internal logic clock of the receiving chip, thus enabling clock domain isolation for both the transmitting and receiving ends.

[0034] Power and ground signals are led out to the packaging substrate through through silicon vias (TSVs) and bump arrays within the silicon interposer.

[0035] In one implementation, the lateral interconnect pathways in the silicon interposer are interconnect pathways between multiple layers of metal wiring, supporting more parallel channels and more complex interconnect topologies; the hybrid bonding pad spacing is no more than 10 μm, and hundreds to thousands of parallel signal lines can be arranged at the edge of the die communication area to provide high total bandwidth in a parallel manner.

[0036] Compared to existing solutions that rely on SerDes for serial data transmission, this embodiment directly transmits high-speed data signals in parallel by setting up a transmit FIFO, transmit circuit, receive circuit, and receive FIFO within each die. This eliminates the need for multi-level processing such as serialization, encoding, CDR recovery, and deserialization, which simplifies the physical layer circuit structure and reduces circuit complexity.

[0037] Understandably, since different protocols have different requirements for output swing, pre-emphasis intensity and common-mode voltage, the swing of traditional CMOS buffer is fixed and the pre-emphasis cannot be adjusted after fabrication. Multiple sets of transmitter circuits are required to be compatible with multiple protocols.

[0038] See Figure 2 In some embodiments, the transmitting circuit includes a main drive branch, a first pre-emphasis branch, a first programmable resistor R0, a second programmable resistor R2, a first operational amplifier U1, a second operational amplifier U2, a first load transistor PM0, a second load transistor NM0, and a feedback resistor R3; the output terminal of the transmitting circuit is a VPAD pad.

[0039] The output of the main drive branch is connected to the output of the first pre-emphasis branch via the first programmable resistor R0, and the output of the first pre-emphasis branch is connected to the output of the transmitting circuit VPAD via the second programmable resistor R2.

[0040] Specifically, the first programmable resistor R0 is connected in series between the output terminal of the main drive branch and the output terminal of the first pre-emphasis branch; the second programmable resistor R2 is connected in series between the output terminal of the first pre-emphasis branch and the output terminal VPAD of the transmitting circuit.

[0041] The non-inverting input of the first operational amplifier U1 is used to connect to the first reference voltage VREF+, and its inverting input is connected to the output terminal VPAD of the transmitting circuit via the feedback resistor R3. The output terminal of the first operational amplifier is connected to the gate of the first load transistor PM0. The non-inverting input of the second operational amplifier U2 is used to connect to the second reference voltage VREF-, and the inverting input of the second operational amplifier is connected to the output terminal VPAD of the transmitting circuit. Its output terminal is connected to the gate of the second load transistor NM0.

[0042] Understandably, the first operational amplifier U1 feeds back the voltage of the output terminal VPAD of the transmitting circuit to the inverting input terminal through the feedback resistor R3, and compares it with the first reference voltage VREF+ at the non-inverting input terminal, thus forming a negative feedback closed loop; the second operational amplifier U2 feeds back the voltage of the output terminal VPAD of the transmitting circuit to the inverting input terminal, and compares it with the second reference voltage VREF- at the non-inverting input terminal, also forming a negative feedback closed loop; wherein, the feedback resistor R3 can be 100kΩ.

[0043] The first load transistor PM0 is connected between the power supply terminal VDDIO and the first common source node of the main drive branch and the first pre-emphasis branch; the second load transistor NM0 is connected between the second common source node of the main drive branch and the first pre-emphasis branch and the ground terminal VSSIO.

[0044] Specifically, the first load transistor PM0 is a PMOS transistor, with its source connected to the power supply terminal VDDIO, its drain connected to the first common-source node, and its gate connected to the output terminal of the first operational amplifier U1; the second load transistor NM0 is an NMOS transistor, with its drain connected to the second common-source node, its source connected to the ground terminal VSSIO, and its gate connected to the output terminal of the second operational amplifier U2.

[0045] Continue reading Figure 2 In some embodiments, the main drive branch includes a first PMOS transistor PM1 and a first NMOS transistor NM1.

[0046] In this circuit, the drain of the first PMOS transistor PM1 is connected to the drain of the first NMOS transistor NM1, serving as the output terminal of the main drive branch; the gate of the first PMOS transistor PM1 is connected to the gate of the first NMOS transistor NM1, which is connected to the data input terminal VIN, and the data input terminal VIN is connected to the output of the transmitting terminal FIFO.

[0047] The first pre-emphasis branch is composed of the second PMOS transistor PM3 and the second NMOS transistor NM3; the drain of the second PMOS transistor PM3 and the drain of the second NMOS transistor NM3 are connected together, serving as the output terminal of the first pre-emphasis branch; the gate of the second PMOS transistor PM3 and the gate of the second NMOS transistor NM3 are connected together to the control terminal VIN-EMP of the first pre-emphasis branch, used to receive the pre-emphasis control signal.

[0048] Specifically, the source of the first PMOS transistor PM1 and the source of the second PMOS transistor PM3 are connected to the first common-source node, which is the drain of the first load transistor PM0; the source of the first NMOS transistor NM1 and the source of the second NMOS transistor NM3 are connected to the second common-source node, which is the drain of the second load transistor NM0.

[0049] In some embodiments, the first programmable resistor R0 and the second programmable resistor R2 are digitally programmable adjustable resistors; the first programmable resistor R0 and the second programmable resistor R2 each include a plurality of parallel resistor units, each resistor unit is connected in series with a switching transistor, and the switching transistor is controlled by a register configuration signal to switch the corresponding resistor unit on or off.

[0050] Specifically, a MOS switch is connected in parallel with each resistor unit. The switch is turned on or off by a register to connect or bypass the resistor unit, thereby combining different total resistance values.

[0051] Understandably, since the resistance values ​​of the first programmable resistor R0 and the second programmable resistor R2 can both be configured digitally via registers, the resistor configuration can be switched between different protocols without modifying the circuit layout, which is beneficial for improving the portability of the transmitting circuit.

[0052] In some embodiments, the control terminal VIN-EMP of the first pre-emphasis branch is configured to turn on the second PMOS transistor PM3 and the second NMOS transistor NM3 when the data transition edge arrives, and to turn off when the data is in a steady state.

[0053] The signal output from the main drive branch is transmitted to the output terminal VPAD of the transmitting circuit via the first programmable resistor R0 and the second programmable resistor R2; the signal output from the first pre-emphasis branch is transmitted to the output terminal VPAD of the transmitting circuit via the second programmable resistor R2, which is used to adjust the pre-emphasis signal enhancement amplitude at the output terminal of the transmitting circuit at the data transition edge; the pre-emphasis signal enhancement amplitude is determined by the resistance ratio of the first programmable resistor to the second programmable resistor.

[0054] See Figures 3 to 6Specifically, the signal output from the main drive branch needs to pass through two stages of resistors, the first programmable resistor R0 and the second programmable resistor R2, to reach the VPAD. However, the signal output from the first pre-emphasis branch only needs to pass through the second programmable resistor R2 to reach the VPAD, which can reduce the voltage drop attenuation through the first programmable resistor R0. When the data transition edge arrives, the pre-emphasis control signal VIN-EMP controls the first pre-emphasis branch to be turned on briefly, directly injecting or pulling current into the drain node of the pre-emphasis branch, generating a larger transition amplitude on the VPAD than the steady state through R2, forming a pre-emphasis pulse. In the steady state, VIN-EMP turns off the first pre-emphasis branch, and only the main drive branch outputs a normal swing signal through the first programmable resistor R0 and the second programmable resistor R2.

[0055] Understandably, increasing the value of R0 or decreasing the value of R2 can enhance the pre-emphasis signal boost, and vice versa. Since both can be dynamically configured via registers, the pre-emphasis signal boost can be adjusted online, adapting to different channel loss conditions such as short traces, long traces, and vias, as well as the pre-emphasis level requirements of different protocols. This allows for switching pre-emphasis configurations between different protocols without modifying the circuit layout, enabling the transmit (TX) architecture to be compatible with the output template requirements of multiple high-speed protocols, thereby reducing repetitive design costs.

[0056] To aid in understanding the technical solutions and their effects provided in the embodiments of this application, the following is combined with... Figures 3 to 6 And see Figure 7 As shown, the transmitting circuit goes through the following four stages in sequence during data transmission; Figure 7 The transient waveforms of the data input terminal VIN, the control terminal VIN-EMP of the first pre-emphasis branch, and the output terminal VPAD of the transmitting circuit are shown below. Figures 3 to 6 ,and Figure 7 The signal states at each stage are explained in the four stages: See Figure 3 In the first stage, i.e., the pull-down discharge stage, the second NMOS transistor NM3 of the first pre-emphasis branch is turned on. Current flows from the output terminal VPAD of the transmitting circuit through the second programmable resistor R2 and the second NMOS transistor NM3, and then through the second load transistor NM0 to the ground terminal VSSIO; at this time, the main drive branch is turned off; corresponding to Figure 7 When the data input terminal VIN transitions from high to low, the control terminal VIN-EMP of the first pre-emphasis branch generates a pulse, and the output terminal VPAD of the transmitting circuit is quickly pulled down, forming a down-pulse pulse lower than the steady-state low level.

[0057] See Figure 4In the second stage, the main drive pull-up stage, the first PMOS transistor PM1 of the main drive branch is turned on. Current flows from the power supply terminal VDDIO through the first load transistor PM0 and the first PMOS transistor PM1, and then through the first programmable resistor R0 and the second programmable resistor R2 to the output terminal VPAD of the transmitting circuit; at this time, the first pre-emphasis branch is turned off; corresponding to Figure 7 During the period when the data input terminal VIN transitions from low to high and the control terminal VIN-EMP of the first pre-emphasis branch is turned off, the output terminal VPAD of the transmitting circuit remains stable at a high level.

[0058] See Figure 5 In the third stage, the pre-emphasis pull-up stage, the second PMOS transistor PM3 of the first pre-emphasis branch is turned on. Current flows from the power supply terminal VDDIO through the first load transistor PM0 and the second PMOS transistor PM3, and then through the second programmable resistor R2 to the output terminal VPAD of the transmitting circuit. At this time, the main drive branch is turned off, and the current does not pass through the first programmable resistor R0. Figure 7 When the data input terminal VIN transitions from low to high, the control terminal VIN-EMP of the first pre-emphasis branch generates a pulse, and the output terminal VPAD of the transmitting circuit is quickly pulled up, forming an overshoot pulse higher than the steady-state high level.

[0059] See Figure 6 In the fourth stage, the main drive pull-down stage, the first NMOS transistor NM1 of the main drive branch is turned on. Current flows from the output terminal VPAD of the transmitting circuit through the second programmable resistor R2, the first programmable resistor R0, and the first NMOS transistor NM1, and then through the second load transistor NM0 to the ground terminal VSSIO; at this time, the first pre-emphasis branch is turned off; corresponding to Figure 7 During the period when the data input terminal VIN transitions from high to low and the control terminal VIN-EMP of the first pre-emphasis branch is turned off, the output terminal VPAD of the transmitting circuit remains stable at a low level.

[0060] It should be noted that in the second and fourth stages, the current of the main drive branch passes through two levels of resistors: the first programmable resistor R0 and the second programmable resistor R2. In the third stage, the current of the first pre-emphasis branch only passes through the second programmable resistor R2 and does not pass through the first programmable resistor R0. As a result, when the data transition edge arrives, the current path impedance of the first pre-emphasis branch is smaller, which can generate a larger transition amplitude at the output terminal VPAD of the transmitting circuit, thus forming a pre-emphasis effect.

[0061] It is understood that in this embodiment, the transmitting circuit operates alternately through the above four stages. When the data transition edge arrives, the first pre-emphasis branch directly drives the output terminal VPAD of the transmitting circuit through the second programmable resistor R2. See [link to relevant documentation]. Figure 7In waveform, this is represented by a pre-emphasis pulse with increased amplitude at the transition edge; this pre-emphasis pulse pre-compensates for the high-frequency loss of the transmission channel, thereby improving the rising and falling edges of the receiver signal and increasing the eye diagram opening of the receiver signal.

[0062] In some embodiments, the first operational amplifier is further configured to adjust the on-resistance of the first load transistor PMO via negative feedback, so that the high level of the output terminal VPAD of the transmitting circuit tracks the first reference voltage VREF+; the second operational amplifier is further configured to adjust the on-resistance of the second load transistor NM0 via negative feedback, so that the low level of the output terminal VPAD of the transmitting circuit tracks the second reference voltage VREF-.

[0063] The output swing of the transmitting circuit is the difference between the high and low levels, and the output swing is adjusted by setting the first reference voltage VREF+ and the second reference voltage VREF-.

[0064] Specifically, the inverting input of the first operational amplifier is connected to the output of the transmitting circuit, VPAD, via feedback resistor R3; the first operational amplifier compares the high level VOH of VPAD with the first reference voltage VREF+, and adjusts the on-resistance of the first load transistor PM0 through negative feedback so that VOH tracks VREF+.

[0065] The inverting input of the second operational amplifier is connected to the output VPAD of the transmitting circuit. The second operational amplifier compares the low level VOL of VPAD with the second reference voltage VREF-, and adjusts the on-resistance of the second load transistor NM0 through negative feedback so that VOL tracks VREF-.

[0066] It is understood that in this embodiment, by configuring the voltage values ​​of VREF+ and VREF- respectively, the high and low output levels can be freely defined without modifying the circuit layout, thereby precisely controlling the output swing to be compatible with different data transmission protocol requirements.

[0067] In some other embodiments, high-speed data signals enter the silicon interposer vertically via hybrid bonding pads, are transmitted through the lateral interconnect pathways inside the silicon interposer to the hybrid bonding pads below the adjacent die, and are then transmitted vertically to the adjacent die; power signals and ground signals are led out to the packaging substrate through silicon vias and bump arrays in the silicon interposer.

[0068] Specifically, the voltage signal output by the transmitting circuit descends vertically through the hybrid bonding pads into the silicon interposer, is transmitted through the lateral interconnect path inside the interposer to the bottom of the opposite die, and then ascends vertically through the hybrid bonding pads to reach the opposite die.

[0069] Optionally, the lateral interconnect paths in the silicon interposer can be interconnect paths between multiple layers of metal wiring, supporting more parallel channels and more complex interconnect topologies. Generally, the spacing between hybrid bonding pads can be no more than 10μm, and the line width can be increased to 1-2μm. Hundreds to thousands of parallel signal lines can be arranged at the edge of the die communication area to provide high total bandwidth in parallel. Since the lateral interconnect distance between dies through the silicon interposer is extremely short, the signal attenuation is small. Therefore, signal integrity can be guaranteed without complex equalization circuits.

[0070] In some embodiments, the transmitting circuit further includes at least one second pre-emphasis branch; the at least one second pre-emphasis branch is connected to the output terminal VPAD of the transmitting circuit via a corresponding third programmable resistor.

[0071] In this embodiment, by adding a second pre-emphasis branch and using a resistor network with different resistance values, multiple levels of pre-emphasis adjustment can be achieved to adapt to transmission channels with longer distances or greater losses. Of course, more pre-emphasis drive branches can be added as needed.

[0072] In some embodiments, the transmitting circuit further includes an eye diagram monitoring circuit or a bit error detection circuit, used to adjust the pre-emphasis signal enhancement amplitude and output swing based on the signal quality feedback from the receiving end.

[0073] In this embodiment, by inserting an eye diagram monitoring or bit error detection circuit on-chip, the pre-emphasis signal enhancement amplitude and output swing can be automatically adjusted according to the signal quality feedback from the receiving end, which can achieve self-balancing and help ensure the signal integrity of the link under various operating conditions.

[0074] Example 2 This application also provides a data transmission method for a semiconductor package, applicable to any of the semiconductor packages described in Embodiment 1 above; such as Figure 8 As shown, the method includes: S110, high-speed data signals, power signals, ground signals and control signals are interconnected between the upper surface of the silicon interposer and at least two dies via hybrid bonding pads.

[0075] Specifically, high-speed data signals, power signals, ground signals, and control signals between the die and the silicon interposer are directly bonded and interconnected via hybrid bonding pads.

[0076] S120. Within each die, the high-speed data signal is transmitted in parallel between two dies via a data transmission link, and the control signal is transmitted via the hybrid bonding pad.

[0077] S130, The power signal and the ground signal are led out to the packaging substrate through the silicon vias and bump array in the silicon interposer.

[0078] In some embodiments, transmitting the high-speed data signal in parallel between two dies via a data transmission link within each die includes: The output signal of the main drive branch is transmitted to the output terminal of the first pre-emphasis branch via the first programmable resistor, and then transmitted to the output terminal of the transmitting circuit via the second programmable resistor. The on-resistance of the first load transistor is adjusted by the first operational amplifier, and the on-resistance of the second load transistor is adjusted by the second operational amplifier.

[0079] In some embodiments, the output signal through the main drive branch includes: When the data transition edge arrives, the first pre-emphasis branch is controlled to be turned on, so that the output signal of the first pre-emphasis branch is transmitted to the output terminal of the transmitting circuit through the second programmable resistor. When the data is in a steady state, the first pre-emphasis branch is turned off, so that the output signal of the main drive branch is transmitted to the output terminal of the transmitting circuit through the first programmable resistor and the second programmable resistor. By adjusting the resistance ratio of the first programmable resistor to the second programmable resistor, the pre-emphasis signal enhancement amplitude at the output of the transmitting circuit at the data transition edge is adjusted.

[0080] In some other embodiments, adjusting the on-resistance of the first load transistor via the first operational amplifier includes: adjusting the on-resistance of the first load transistor via negative feedback through the first operational amplifier, so that the high level at the output of the transmitting circuit is equal to the first reference voltage. The step of adjusting the on-resistance of the second load transistor via the second operational amplifier includes: adjusting the on-resistance of the second load transistor via negative feedback through the second operational amplifier, so that the low level at the output of the transmitting circuit is equal to the second reference voltage; and adjusting the output swing of the transmitting circuit by configuring the first reference voltage and the second reference voltage.

[0081] In some embodiments, the high-speed data signal interconnected between the upper surface of the silicon interposer and at least two dies via the hybrid bonding pad includes: vertically transmitting the high-speed data signal into the silicon interposer via the hybrid bonding pad; transmitting the high-speed data signal to the hybrid bonding pad below the adjacent die via a lateral interconnect path inside the silicon interposer; and then vertically transmitting the high-speed data signal to the adjacent die via the hybrid bonding pad.

[0082] As an optional embodiment, the high-speed data signal is transmitted between two dies in parallel via a data transmission link within each die, including: S121, writing data generated by the internal logic of the transmitting die into the transmitting end FIFO.

[0083] The transmitting FIFO achieves cross-clock domain isolation, decoupling the internal logic clock of the transmitting chip from the interconnect transmission clock, while providing a data buffer to smooth data rate fluctuations; the transmitting FIFO writes data with the internal logic clock of the transmitting chip and reads data with the transmission clock.

[0084] S122. The data output from the FIFO at the transmitting end is sent to the transmitting circuit, and the transmitting circuit outputs a voltage signal.

[0085] The transmitting circuit includes a main drive branch, a first pre-emphasis branch, a first programmable resistor R0, a second programmable resistor R2, a first operational amplifier, a second operational amplifier, a first load transistor PM0, and a second load transistor NM0. The signal output from the main drive branch is transmitted to the output terminal VPAD of the transmitting circuit via the first programmable resistor R0 and the second programmable resistor R2. The signal output from the first pre-emphasis branch is transmitted to the output terminal VPAD of the transmitting circuit only via the second programmable resistor R2.

[0086] S123. Control the first pre-emphasis branch to be turned on when the data transition edge arrives, and to be turned off when the data is in a steady state.

[0087] Specifically, when the data transition edge arrives, the pre-emphasis control signal controls the first pre-emphasis branch to be briefly turned on, and current is injected or pulled out at the drain node of the pre-emphasis branch. This generates a larger transition amplitude than the steady state at the output terminal VPAD of the transmitting circuit through the second programmable resistor R2, forming a pre-emphasis pulse. In the steady state, the pre-emphasis control signal turns off the first pre-emphasis branch, and only the main drive branch outputs the normal swing signal through the first programmable resistor R0 and the second programmable resistor R2.

[0088] S124. Through the negative feedback closed-loop control of the first operational amplifier and the second operational amplifier, the high level of the output terminal VPAD of the transmitting circuit tracks the first reference voltage VREF+, and the low level tracks the second reference voltage VREF-.

[0089] S125. The voltage signal is vertically transmitted into the silicon interposer via the hybrid bonding pads, then transmitted to the receiving die via the lateral interconnect path inside the silicon interposer, and then vertically transmitted to the receiving die via the hybrid bonding pads.

[0090] In some embodiments, during data transmission, the method further includes: adjusting the output swing of the output terminal VPAD of the transmitting circuit and the pre-emphasis signal enhancement amplitude by configuring the first reference voltage VREF+, the second reference voltage VREF-, and the resistance ratio of the first programmable resistor R0 to the second programmable resistor R2, so as to be compatible with the requirements of different protocols.

[0091] Since the technical concepts are basically the same for all embodiments, the relevant parts can be referred to the description of Embodiment 1.

[0092] The foregoing has described exemplary embodiments of this specification. It should be understood that in some cases, the modules described in this specification may be divided in a manner different from that in the embodiments, and the described actions or steps may be performed in a different order than that in the embodiments, while still achieving the desired result. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0093] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A semiconductor package, comprising: The package substrate, a silicon interposer layer located above the package substrate, and at least two dies; characterized in that the silicon interposer layer is connected to the package substrate via a bump array, and the at least two dies are connected to the upper surface of the silicon interposer layer via hybrid bonding pads. High-speed data signals, power signals, ground signals, and control signals between the die and the silicon interposer are directly bonded and interconnected via hybrid bonding pads. Each die is equipped with a data transmission link, which includes a transmit first-in-first-out queue, a transmit circuit, a receive circuit, and a receive first-in-first-out queue. The high-speed data signal is transmitted in parallel between the transmit circuit and the receive circuit of the two dies via the data transmission link.

2. The semiconductor package according to claim 1, characterized in that, The transmitting circuit includes a main drive branch, a first pre-emphasis branch, a first programmable resistor, a second programmable resistor, a first operational amplifier, a second operational amplifier, a first load transistor, a second load transistor, and a feedback resistor; The output terminal of the main drive branch is connected to the output terminal of the first pre-emphasis branch via the first programmable resistor, and the output terminal of the first pre-emphasis branch is connected to the output terminal of the transmitting circuit via the second programmable resistor. The non-inverting input of the first operational amplifier is used to connect to the first reference voltage, the inverting input of the first operational amplifier is connected to the output of the transmitting circuit through the feedback resistor, and the output of the first operational amplifier is connected to the gate of the first load transistor. The non-inverting input of the second operational amplifier is used to connect to the second reference voltage, the inverting input of the second operational amplifier is connected to the output of the transmitting circuit, and the output of the second operational amplifier is connected to the gate of the second load transistor. The first load transistor is connected between the power supply terminal and the first common source node of the main drive branch and the first pre-emphasis branch, and the second load transistor is connected between the second common source node of the main drive branch and the first pre-emphasis branch and the ground terminal.

3. The semiconductor package according to claim 2, characterized in that, The main drive branch includes: a first PMOS transistor and a first NMOS transistor, wherein the drains of the first PMOS transistor and the first NMOS transistor are connected together, and the gates of the first PMOS transistor and the first NMOS transistor are connected together at the data input terminal; the first pre-emphasis branch is composed of a second PMOS transistor and a second NMOS transistor; the drains of the second PMOS transistor and the second NMOS transistor are connected together; and the gates of the second PMOS transistor and the second NMOS transistor are connected together at the control terminal of the first pre-emphasis branch.

4. The semiconductor package according to claim 2, characterized in that, The first programmable resistor and the second programmable resistor are digitally programmable and adjustable resistors. The first programmable resistor and the second programmable resistor each include multiple resistor units connected in parallel. Each resistor unit is connected in series with a switching transistor. The switching transistor is controlled by a register configuration signal to switch the corresponding resistor unit on or off.

5. The semiconductor package according to claim 3, characterized in that, The control terminal of the first pre-emphasis branch is configured to control the second PMOS transistor and the second NMOS transistor to turn on when the data transition edge arrives, and to turn off when the data is in a steady state. The signal output from the main drive branch is transmitted to the output of the transmitting circuit via the first programmable resistor and the second programmable resistor. The signal output from the first pre-emphasis branch is transmitted to the output of the transmitting circuit via the second programmable resistor, which is used to adjust the pre-emphasis signal enhancement amplitude at the output of the transmitting circuit at the data transition edge. The signal enhancement amplitude is determined by the resistance ratio of the first programmable resistor to the second programmable resistor.

6. The semiconductor package according to claim 2, characterized in that, The first operational amplifier is further configured to adjust the on-resistance of the first load transistor via negative feedback, so that the high level at the output of the transmitting circuit is equal to the first reference voltage; The second operational amplifier is further configured to adjust the on-resistance of the second load transistor via negative feedback, so that the low level at the output of the transmitting circuit is equal to the second reference voltage; The output swing of the transmitting circuit is the difference between the high level and the low level, and the output swing is adjusted by setting the first reference voltage and the second reference voltage.

7. The semiconductor package according to claim 1, characterized in that, The high-speed data signal enters the silicon interposer vertically through the hybrid bonding pads, is transmitted through the lateral interconnect pathways inside the silicon interposer to the hybrid bonding pads below the adjacent die, and is then transmitted vertically to the adjacent die; the power signal and the ground signal are led out to the packaging substrate through the silicon vias and the bump array in the silicon interposer.

8. The semiconductor package according to claim 2, characterized in that, The transmitting circuit further includes at least one second pre-emphasis branch, and at least one second pre-emphasis branch is connected to the output terminal of the transmitting circuit via a corresponding third programmable resistor.

9. The semiconductor package according to claim 2, characterized in that, The transmitting circuit also includes an eye diagram monitoring circuit or a bit error detection circuit, used to adjust the pre-emphasis signal enhancement amplitude and output swing based on the signal quality feedback from the receiving end.

10. A data transmission method for a semiconductor package, characterized in that, include: High-speed data signals, power signals, ground signals, and control signals are interconnected between the upper surface of the silicon interposer and at least two dies via hybrid bonding pads. Within each die, the high-speed data signal is transmitted in parallel between two dies via a data transmission link, and the control signal is transmitted via the hybrid bonding pads; In addition, the power signal and the ground signal are led out to the packaging substrate through silicon vias and bump arrays in the silicon interposer.

11. The data transmission method according to claim 10, characterized in that, The transmission of the high-speed data signal between two dies in parallel via a data transmission link within each die includes: The output signal of the main drive branch is transmitted to the output terminal of the first pre-emphasis branch via the first programmable resistor, and then transmitted to the output terminal of the transmitting circuit via the second programmable resistor. The on-resistance of the first load transistor is adjusted by the first operational amplifier, and the on-resistance of the second load transistor is adjusted by the second operational amplifier.

12. The data transmission method according to claim 11, characterized in that, The signal output through the main drive branch includes: When the data transition edge arrives, the first pre-emphasis branch is controlled to be turned on, so that the output signal of the first pre-emphasis branch is transmitted to the output terminal of the transmitting circuit through the second programmable resistor. When the data is in a steady state, the first pre-emphasis branch is turned off, so that the output signal of the main drive branch is transmitted to the output terminal of the transmitting circuit through the first programmable resistor and the second programmable resistor. By adjusting the resistance ratio of the first programmable resistor to the second programmable resistor, the pre-emphasis signal enhancement amplitude at the output of the transmitting circuit at the data transition edge is adjusted.

13. The data transmission method according to claim 11, characterized in that, The step of adjusting the on-resistance of the first load transistor through the first operational amplifier includes: adjusting the on-resistance of the first load transistor through the first operational amplifier via negative feedback, so that the high level at the output of the transmitting circuit is equal to the first reference voltage; The step of adjusting the on-resistance of the second load transistor via the second operational amplifier includes: adjusting the on-resistance of the second load transistor via negative feedback through the second operational amplifier, so that the low level at the output of the transmitting circuit is equal to the second reference voltage; The output swing of the transmitting circuit is adjusted by configuring the first reference voltage and the second reference voltage.

14. The data transmission method according to claim 10, characterized in that, The high-speed data signal interconnected between the hybrid bonding pads on the upper surface of the silicon interposer and at least two grains includes: vertically transmitting the high-speed data signal into the silicon interposer via the hybrid bonding pads; The high-speed data signal is transmitted to the hybrid bonding pads below the adjacent die via the lateral interconnect path inside the silicon interposer, and then the high-speed data signal is vertically transmitted to the adjacent die via the hybrid bonding pads.